there's this paper on "parallel FFT hashing" which is written in words but makes absolutely no sense to me https://link.springer.com/content/pdf/10.1007/3-540-58108-1_18.pdf also djb glazes up schnorr too much so i just assume he sucks
6
Replies
0
Boosts
@hipsterelectron
> parallel
Okay, with you so far.
> FFT
Fast Fourier Transform, uh huh, cool math shit translates signals from time to frequency domain—we’ve worked together …
> hashing
What the fuck? What does- How- huh?
@c0dec0dec0de there's actually a lot of cool stuff you can do with real number digit expansions
@hipsterelectron I mean, aren’t the DCT and FFT at least partially reversible? Dependent on knowing your parameters, sure, but like hashing should (generally) not be based on reversible functions because they’re supposed to be one-way. It’s not the same as encryption where you want a trap-door function that’s apparently one-way unless you have the key (parameters)…
@hipsterelectron you’re binning information, but you can recover a time domain signal (without some corruption from discretization, obvs) from an FFTed frequency domain representation.
@c0dec0dec0de that's a great point regarding binning. the digit expansion approaches give you actual digits of e.g. pi i believe so i do wonder whether they could be fruitfully obscured
